Multimodal Machine Learning and Data Fusion for Environmental Systems
Multimodal Machine Learning and Data Fusion for Environmental Systems
This graduate fellowship applies multimodal machine learning (ML) and data-fusion techniques to integrate diverse environmental observations—including geophysical, geochemical, remote-sensing, and sensor datasets—to improve site characterization, monitoring, and predictive understanding of complex subsurface systems.
Fellows will develop and apply ML methods to learn relationships across data modalities, extract meaningful patterns, detect anomalies, and integrate field measurements, imagery, and simulation outputs. Research activities may include data preparation and analysis, representation learning, time-series modeling, model development and validation, and evaluation of uncertainty in environmental predictions. This work will strengthen U.S. Department of Energy Office of Environmental Management (DOE-EM) capabilities for environmental modeling, risk assessment, data-informed decision-making, long-term stewardship, and cleanup optimization.
Appointment Details: This is a full-time, 13-month graduate fellowship assignment. The anticipated start date shown in the posting is an example and may be adjusted based on program needs, candidate availability, and onboarding requirements.
Desired Skills: Multimodal machine learning and data fusion; integration and analysis of geophysical, geochemical, remote-sensing, and sensor data; representation learning, time-series modeling, and anomaly detection; Python programming; and experience with machine-learning frameworks, such as PyTorch.
Desired Education: Current graduate student in Environmental Engineering, Geosciences, Computer Science, Data Science, Earth and Environmental Science, or a closely related field.